Back

Mechanistic machine learning enables interpretable and generalizable prediction of prime editing outcomes

Hsu, A.; Chen, P. J.; Li, A. H.; Hemez, C. F.; Gao, X. D.; Terrey, M.; Nelson, C.; Selvam, V.; Cristian, A.; McElroy, A. N.; Steinbeck, B. J.; Mahadeshwar, G. K.; Pandey, S.; Barsdale, Z.; Chen, P. Z.; Sousa, A. A.; Sakai, H. A.; Silverstein, R. A.; Morad, I.; Krueger, R. K.; Shen, M. W.; Kleinstiver, B. P.; Lutz, C. M.; Tolar, J.; Blazar, B. R.; Osborne, M.; Liu, D.

2026-02-20 biochemistry
10.64898/2026.02.20.706353 bioRxiv
Show abstract

Although prime editing (PE) can effect virtually any specified local change to genomic DNA in living systems, its efficient application currently requires extensive optimization of prime editing guide RNA (pegRNA) sequences. We present OptiPrime, a machine learning model of PE efficiency based on our current understanding of the mechanism of prime editing. OptiPrime achieves state-of-the-art accuracy on PE efficiency prediction and also enables prediction of nicking guide RNA (PE3) and dual pegRNA (twinPE) outcomes. We validated that OptiPrime has learned the determinants of mammalian mismatch repair (MMR), and is therefore well suited for nominating MMR-evasive silent edits that improve PE efficiency. We demonstrate the utility of OptiPrime in a variety of prospective therapeutic contexts, including in primary human and mouse cells. Finally, we show how OptiPrime can be used to achieve highly streamlined and efficient in vivo correction of a pathogenic mutation in the brain of a mouse model of KIF1A-associated neurological disorder.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

1
Nature Communications
○ 5641 papers in training set
Top 4%
26.7%
Similar papers in this journal
2
Nature Biotechnology
○ 172 papers in training set
Top 0.3%
7.9%
Similar papers in this journal
3
Nature Methods
○ 385 papers in training set
Top 1%
7.3%
Similar papers in this journal
4
Nucleic Acids Research
◐ 1281 papers in training set
Top 2%
7.3%
Similar papers in this journal
5
Nature
○ 645 papers in training set
Top 3%
6.3%
Similar papers in this journal
50% of probability mass above
6
Molecular Cell
○ 350 papers in training set
Top 1%
5.5%
7
Cell
○ 431 papers in training set
Top 1%
5.5%
8
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 19%
2.8%
9
Nature Chemical Biology
○ 119 papers in training set
Top 1.0%
2.7%
10
eLife
● 5828 papers in training set
Top 42%
2.4%
11
Science
● 477 papers in training set
Top 4%
2.4%
12
Nature Biomedical Engineering
○ 47 papers in training set
Top 0.7%
1.9%
13
Nature Structural & Molecular Biology
236 papers in training set
Top 2%
1.5%
14
Cell Systems
○ 201 papers in training set
Top 3%
1.5%
15
Cell Reports
○ 1498 papers in training set
Top 21%
1.3%
16
Science Advances
● 1243 papers in training set
Top 23%
1.3%
17
Genome Biology
○ 637 papers in training set
Top 7%
1.1%
18
Advanced Science
○ 286 papers in training set
Top 9%
0.8%
19
Nature Genetics
○ 286 papers in training set
Top 5%
0.8%
20
ACS Central Science
● 71 papers in training set
Top 1%
0.8%
21
Nature Cancer
○ 39 papers in training set
Top 2%
0.6%
22
Cell Reports Methods
○ 165 papers in training set
Top 4%
0.6%
23
Communications Biology
○ 993 papers in training set
Top 34%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.